Recently (since June 9, 2026), developers of neural networks have encountered a phenomenon that could have far-reaching consequences. Manufacturers of LLM are not only explicitly refusing clients to work on certain topics, but also implementing implicit mechanisms. The well-known model Claude, which is considered one of the best or even the best LLM for software development, quietly and imperceptibly (unlike issues such as cybersecurity, biology, and chemistry) slows down when it detects attempts to use it for developing other models.
It is officially stated in the 'model card' for version 5 of Claude Mythos and Claude Fable on page 13 of the proprietary PDF:
In light of the ability of modern models to accelerate their own development, we have implemented new measures that limit the effectiveness of Claude for requests aimed at developing advanced machine learning models (for example, creating pre-training pipelines, distributed training infrastructure, or designing machine learning accelerators). Using Claude to develop competing models already violates our terms of service, but enforcing this limitation through our protective measures helps prevent accelerating the actions of those most likely to breach these terms.
Unlike our protective measures in cybersecurity, biology, and chemistry, as well as attempts at distillation, these protective measures will not be visible to the user. Fable 5 will not switch to another model. Instead, the safeguards will limit effectiveness through methods such as prompt modifications, guiding vectors, or parameter-efficient fine-tuning (PEFT). These protective measures will not affect the overwhelming majority of programming work. We estimate that they will impact ~0.03% of traffic, concentrated in less than 0.1% of organizations.
Discussion on Y combinator
Source: linux.org.ru
